HCI Lesson 16 - Info Search and Visualization Total by xiangpeng

VIEWS: 7 PAGES: 41

									                                                                                 1

               Information Search and Visualization

   Information Terminology
      Information Retrieval
      Information gathering, seeking, filtering, and visualization
      Task objects: e.g., video clips, documents
      Task actions: browsing and searching
      Interface actions: Scrolling, joining, zooming, linking
      Database Management – refers to structured relational database systems,
       well defined attributes and sort-keys
      Data mining, data warehouses, data marts
      Knowledge networks, semantic webs
                                                                               2

               Information Search and Visualization

   Information Terminology
      Specific fact finding: known-item search
           • Example: find the email address of Keith Jackson
      Extended fact finding
           • Example: What are the sonnets by Shakespeare
      Exploration of availability
           • Example: Is there new work in process control published by IEEE
      Open ended browsing and problem analysis
           • Is there new research on the use of cell phones in China
                                                                                               3

               Information Search and Visualization

   Searching in Text Documents and Database Querying
      Google’s Link Based Ranking Measure – PageRank (Brin & Page, 1998)
           • Computes a query independent score for each document
           • Takes into consideration the importance of the pages that point to a given page
           • The big dogs know where to hunt
      SQL (database query language)
           • Example:
               SELECT DOCUMENT#
               FROM JOURNAL = MY_FAVORITE_JOURNAL
               WHERE (DATE > 2001 AND DATE <= 2003)
                    AND (LANGUAGE = ENGLISH)
                    AND (PUBLISHER = HFES OR ACM)
      Natural Language Queries
           • Mainly just eliminates frequent terms
                                                        4

               Information Search and Visualization

   Searching in Text Documents and Database Querying
      Form-Fillin Queries (http://thomas.loc.gov/)
                                                               5

              Information Search and Visualization

   Searching in Text Documents and Database Querying
      Phases of search
          • Formulation: expressing the search
          • Initiation of action: launching the search
          • Review of results: reading messages and outcomes
          • Refinement: formulating the next step
          • Use: compiling or disseminating information
                                                                                       6

              Information Search and Visualization

   Searching in Text Documents and Database Querying
      Formulation
          • Identify the source of the information (e.g., within a specific library)
          • Use fields to limit the search (e.g., year or language)
          • Recognize phrases to allow entry of names (e.g., Abraham Lincoln)
               – Allow for search my phrase or individual items in the phrase
          • Apply variants to relax the search constraints
               – Case sensitivity (JEFFERSON, Jefferson)
               – Stemming (sing, singing)
               – Partial matches (biology, psychobiology, sociobiology)
               – Phonetic variations (Smith, Smyth, Smythe)
               – Abbreviations (ATT, NCR)
               – Synonyms (West Coast retrieves Washington, Oregon and California)
                                                        7

             Information Search and Visualization

   Searching in Text Documents and Database Querying
      Formulation
                                                                                        8

               Information Search and Visualization

   Searching in Text Documents and Database Querying
      Initiation of Action
           • Explicit initiation (e.g., search button)
           • Implicit initiation: each change to a component of the formulation phase
              immediately produces a new set of search results (e.g., Google)
                                                             9

              Information Search and Visualization

   Searching in Text Documents and Database Querying
      Review of Results
          • Users can read messages and view textual lists
          • Allow the user to control
              – The number of results
              – Which fields are displayed
              – The sequence of the results
              – How results are clustered
                                                        10

              Information Search and Visualization

   Searching in Text Documents and Database Querying
      Review of Results
          • Clustering
                                                        11

              Information Search and Visualization

   Searching in Text Documents and Database Querying
      Review of Results
          • User control
                                                                                         12

              Information Search and Visualization

   Searching in Text Documents and Database Querying
      Refinement
          • In the event of few results, indicate that using fewer search criteria, or
            partial matches may increase the number of hits
          • Suggested spellings




          • If no results are found, always provide users with that information
                                                                                       13

              Information Search and Visualization

   Searching in Text Documents and Database Querying
      Use Results
          • Merge, save, distributed via email, output to visualization programs, or
            statistical tools
                                                                                         14

              Information Search and Visualization

   Multimedia Document Searches
      Most systems used to locate images, video, sound and animation
       depend on metadata
      Example: search of a photo library by date, photographer or text
       captions
          • Requires significant human effort to add captions and annotate
      Image search: query by image content
      Map search
          • Search by latitude and longitude
          • Search by features (e.g., search for all cities in northwest United States
            with airports)
                                                                               15

           Information Search and Visualization

 Picasa
    Supports browse and search of photos in public albums
    Automatically organizes the user’s online photo collection based to who's in
     each picture
                                                                                    16

               Information Search and Visualization

   Other Searching Mechanisms
      Sound Search – Music-information retrieval (MIR)
           • Users can play or sing as input, and matching songs will be returned
      Video Search
           • Segment into scenes
           • Allow scene skipping
      Animation Search
           • Examples: search for morphing faces
                                                                                     17

               Information Search and Visualization

   Video Search
        Informedia
      Designed at CMU to solve the problem of searching huge collections of video
       and audio recordings
      Developed new approaches for automated video and audio indexing,
       navigation, visualization, search
      Provides full-content search and retrieval of current and past TV and radio
       news and documentary broadcasts.
      Generates various summaries for each story segment: headlines, filmstrip
       story-boards and video-skims
                                                                                 18

               Information Search and Visualization

   Video Search - Informedia
      Example: 12 documents returned for "El Niño" query along with different
       multimedia abstractions from certain documents
                                                                                          19

               Information Search and Visualization

   Advanced Filtering and Search Interfaces
      Filtering with complex Boolean queries
           • Example: List all employees who live in Denver and Detroit
           • Would most likely result in a null result since “and” implies intersection
           • Most employees do not live in both locations
           • Other approaches
                – Venn Diagrams
                – Decision Tables
                – Metaphors of water flowing through a series of filters
      Automatic Filtering
           • Selective dissemination of information
           • Filtering email before it is placed in the Inbox
                                                                    20

               Information Search and Visualization

   Advanced Filtering and Search Interfaces
      Dynamic queries
           • Uses direct manipulation objects
           • http://www.bluenile.com/diamond_search.asp?track=dss
                                                         21

               Information Search and Visualization

   Advanced Filtering and Search Interfaces
      Metadata search (e.g., Flamenco)
           • Attribute values are selected by the user
           • http://flamenco.berkeley.edu/demos.html
                                                                                                 22

               Information Search and Visualization

   Advanced Filtering and Search Interfaces
      Collaborative Filtering
           • Users work together to define filtering criteria in large information spaces
           • Example: If you ranked five movies highly, the algorithm provides you with a list
             of other movies that were rated highly by people who liked your five movies
      Visual Searches
           • Examples: Selecting dates on calendars or seats from a plane image
                                                     23

              Information Search and Visualization

   Advanced Filtering and Search Interfaces
                                                                                       24

               Information Search and Visualization

   Information Visualization
      The use of interactive visual representations of abstract data to amplify cognition
      Scientific Visualization: requires two dimensions because typical questions
       involve
           • Continuous variables
           • Volumes
      Information Visualization involve
           • Categorical variables
           • Discovery of patterns
           • Trends
           • Clusters
           • Outliers
           • Gaps in data
                                                                                    25

                 Information Search and Visualization

   Information Visualization
      Uses human perceptual abilities to make discoveries, decisions and propose
       explanations
      Users can scan, recognize and recall images quickly
      Users can detect changes in size, color, shape, movement and texture
      IV Rule
           • Overview first
           • Zoom and filter
           • Details on demand
                                                      26

               Information Search and Visualization

   Information Visualization
      1D Linear Data
           • Text documents, dictionaries
           • Organized sequentially
           • Example: view 4000 lines of
             code
           • Newest lines are in red,
             oldest lines in blue
           • Browser window shows code
             overview and detail window
                                                      27

               Information Search and Visualization

   Information Visualization
      1D Linear Data
           • All the words in Alice in
             Wonderland, arranged in an
             arc, starting at 12:00
           • Lines are drawn around the
             outside, words around the
             inside
           • Words that appear more
             often are brighter
                                                                                   28

               Information Search and Visualization

   Information Visualization
                                              •Proximity indicates similarity of
      2D Map Data
                                              topics
           • Planar data include geographic
                                              •Height reflects the number of
             maps                             documents
           • Each item has task domain
             attributes, (e.g., name)
           • Each item has interface
             features (e.g., size or color)
           • User tasks (find adjacent
             items, regions containing
             items, paths between items
                                                      29

               Information Search and Visualization

   Information Visualization
      3D World Data
           • Real world objects –
             molecules, human body,
             buildings and the relationships
             between the objects
           • Users work with continuous
             variables (e.g., temperature)
                                                                                 30

               Information Search and Visualization

   Information Visualization
                                               •www.inxight.com
      Multidimensional data
                                               •Example of listing of houses for sale
           • Extracted data from statistical   •Spreadsheet metaphor
             databases
           • Tasks include finding patterns,
             correlations between pairs of
             variables, clusters, gaps and
             outliers
                                                                                        31

               Information Search and Visualization

   Information Visualization
                                                 •http://www.cs.umd.edu/hcil/bioinfovis/hce.
      Multidimensional data                     shtml
                                                 •Example: hierarchical clustering of gene
           • Hierarchical or k-means             expression data
             clustering to identify similar      • Identifying clusters of genes that are
                                                 activated with malignant as opposed to
             items                               benign melanoma (skin cancer)

           • Hierarchical: identifies close
             pairs of items and forms ever-
             larger clusters until every point
             is included in the cluster
           • K-means: starts when users
             specify how many clusters to
             create, then the algorithm
             places every item into the most
             appropriate cluster
                                                                              32

               Information Search and Visualization

   Information Visualization
                                           •www.cs.umd.edu/hcil/lifelines
      Temporal Data                       •Example: Patient Medical Record
      Illnesses, Vaccinations,
       Surgeries, Lab Results
      Events have a start/end time,
       and items may overlap
      Tasks: finding all events before,
       after or during some time period
       or moment
                                                                            33

               Information Search and Visualization

   Information Visualization
                                              Example: Organization Chart
      Tree Data
           • Collection of items where each
             item has a link to one parent
             item
                                                                 34

               Information Search and Visualization

   Information Visualization
      Tree Data
           • Hyperbolic Tree Structure
           • Limit the number of nodes in the center of the UI
                                                                            35

               Information Search and Visualization
                                              Example: www.smartmoney.com
                                              Map of the Market
   Information Visualization
      TreeMap
           • Each rectangle represents a
             stock and are organized by
             industry groups
           • The rectangle is proportional
             to the market capitalization
           • The color indicates gain/loss
           • “N” indicates a link to a news
             story
                                                                      36

               Information Search and Visualization

   Information Visualization
                                               Example: ILOG Jviews
      Network Data
           • When items are linked to an
             arbitrary number of other items
           • Users often want to know the
             shortest or least costly path
             connecting two items
                                                 37

          Information Search and Visualization

 Information Visualization
     Parallel Coordinates
                                                38

         Information Search and Visualization

 Star Plots
                                                                                        39

               Information Search and Visualization

   Information Visualization
      Overview Task
           • Users can get a overview of the entire collection
           • Zoom
           • Detail View
           • Movable field-of-view-box
      Filter Task
           • Users can filter-out items that are not of interest
      Details-on-demand Task
           • Users can select an item or group to set details
      Relate Task
           • Users can relate items or groups within a collection
           • Show relationships by proximity, containment, connection or color coding
                                                                40

               Information Search and Visualization

   Information Visualization
      History Task
           • Supports undo, replay and progressive refinement
      Extract Task
           • Allows extraction of sub-collections
           • Send items are obtained
               – Save
               – Email
               – Insert to a statistical package
                                                   41

            Information Search and Visualization

   Periodic table of data visualization methods
   Web Site

								
To top